Structural and functional characterization of the contractile aorta and associated hemocytes of the mosquito Anopheles gambiae

In insects, the open circulatory system consists of the pulsatile organs, the hemocoel, and the hemolymph (Jones, 1977; Chapman et al., 2013; Klowden, 2013). The main pulsatile organ – or pump – is the dorsal vessel, which extends from the head to the posterior of the abdomen and is positioned along the midline of the organism. The dorsal vessel is divided into an aorta in the head and thorax, and a heart in the dorsal abdomen. In Anopheles gambiae adults, the heart contracts in both anterograde and retrograde directions (Glenn et al., 2010). During anterograde contractions, hemolymph enters the lumen of the heart through six pairs of incurrent ostia that are positioned in abdominal segments 2–7 and exits the dorsal vessel through an excurrent opening located in the head (Glenn et al., 2010; League et al., 2015). During retrograde contractions, hemolymph flowing through a pair of venous channels in the thorax enters the heart through an ostial pair positioned at the thoraco-abdominal junction and exits the heart through an excurrent opening located in the eighth abdominal segment (Glenn et al., 2010). The circulation of hemolymph in areas that are distant from the dorsal vessel is aided by accessory pulsatile organs. These auxiliary hearts are located at the base of the antennae, in the scutellum, and in the ventral abdomen (Clements, 1956; Andereck et al., 2010; Boppana and Hillyer, 2014; Chintapalli and Hillyer, 2016).

Although the structure and function of the adult mosquito heart has been described in detail (Glenn et al., 2010; Leódido et al., 2013; League et al., 2015), the physiology of the aorta is unknown. This is also true for other flies – the members of the order Diptera – even though Drosophila melanogaster is often used as a model for human cardiac disease (Choma et al., 2011; Piazza and Wessells, 2011; Ma, 2016). We speculate that the reason behind this gap in knowledge is that the aorta is not very accessible to researchers. That is, whereas the heart is attached to the dorsal midline of the cuticle as it traverses the abdomen, the aorta is embedded within the indirect flight muscles. Furthermore, the thoracic cuticle is often more sclerotized than the abdominal cuticle, and this prevents the visualization of the aorta through the exoskeleton of intact insects.

In this study, we used a combination of resection, fluorescence staining, and still- and video-imaging to describe the structure and location of the aorta of adult A. gambiae. Furthermore, we characterize the functional mechanics of flow through the aorta and the conical chamber of the heart – including the entry of hemolymph through the thoraco-abdominal ostia – and show that the aorta has contractile activity. Finally, we show that hemocytes are present on the outer surface of the aorta but that they are few in number and do not aggregate in response to infection.

Anopheles gambiae Giles sensu stricto (G3 strain) were reared and maintained at 27°C and 75% relative humidity under a 12 h:12 h light:dark photoperiod. Adults were fed a 10% sucrose solution ad libitum and maintained as previously described (Estévez-Lao et al., 2013). Experiments were carried out on adult female mosquitoes at 6 days post-eclosion.

Mosquitoes were anesthetized on ice and dissected under a stereo microscope in order to isolate (i) the entire aorta, (ii) portions of the aorta, or (iii) the aorta and the heart. To initiate the resection procedures, the ventral side of the thorax (including the legs) and a ventral slice of the abdomen were removed by bisecting the mosquito along a coronal plane in alignment with the ventral side of the neck. A slice of the dorsal thorax was then removed along a coronal plane by bisecting the mid-posterior scutum, and the mosquitoes were placed ventral side up on a glass slide in phosphate-buffered saline (PBS) containing 0.1% Tween-20 (Fisher Scientific, Pittsburgh, PA, USA).

To isolate the entire aorta, the alimentary tract was removed, the mosquitoes were bisected along a transverse plane at the thoraco-abdominal junction, and the head and aorta were pulled away from the thorax. The ventral side of the head was then removed in order to expose the pharyngeal plate, and this structure was isolated with the aorta attached. The portion of the aorta containing the anterior aorta and bulbous chamber was isolated as above, except the mosquitoes were bisected along a transverse plane at the junction of the meso- and metathorax. To isolate the portion of the aorta containing the bulbous chamber and the posterior aorta, the head and the alimentary tract were removed, and the aorta was isolated from the thorax. The aorta was separated from the heart by cutting with a fine blade at the anterior junction with the conical chamber. To isolate the aorta with the heart, the head and the alimentary tract were removed, and the heart was detached from the dorsal cuticle by disrupting all alary muscles with an insect pin. The aorta and heart were then removed from the thorax and dorsal cuticle.

For muscle staining, each mosquito was anesthetized on ice and, using a finely pulled glass needle, 0.15–0.20 µl of 16% formaldehyde was injected into the hemocoel through the thoracic anepisternal cleft. After a 5 min incubation period, the aorta was isolated and immersed in 13 µmol l⁻¹ phalloidin-Alexa Fluor 488 (Invitrogen, Carlsbad, CA, USA), 0.16 mmol l⁻¹ Hoechst 33342 (Invitrogen) and 0.1% Triton X-100 (Fisher Scientific) in PBS. The immersed aorta was rocked on a depression slide for 30 min, washed three times in PBS, and mounted in Aqua Poly/Mount (Polysciences, Warrington, PA, USA).

To label hemocytes, Vybrant CM-DiI Cell-Labeling Solution (Invitrogen) – a dye that labels hemocytes in live mosquitoes – was used as previously described (King and Hillyer, 2012). Briefly, mosquitoes were anesthetized and injected with 75 mmol l⁻¹ Vybrant CM-DiI Cell-Labeling Solution and 0.75 mmol l⁻¹ Hoechst 33342 in PBS. Live mosquitoes were incubated at 27°C for 20 min, which allows the dye to be incorporated into the hemocytes. Mosquitoes were anesthetized again and used for either live imaging or aorta resection. To label the muscle of aortas with CM-DiI labeled hemocytes, muscle was stained with phalloidin-Alexa Fluor 488 as described above, with the exception that Triton X-100 was not included in the solution.

Specimens were visualized under brightfield and/or epi-fluorescence illumination using a Nikon 90i compound microscope (Nikon Corporation, Tokyo, Japan) equipped with a Nikon Intensilight C-HGFI fluorescence illumination unit. Image Z-stacks were acquired using a Nikon DS-Qi1Mc CCD camera, a linear encoded Z-motor and Nikon Advanced Research NIS-Elements software. Images were either selected from a layer of the Z-stack or rendered into a two-dimensional image using the Extended Depth of Focus tool of NIS-Elements software.

For intravital video-imaging of hemolymph flow and hemocytes, intact mosquitoes that had been injected with 1 µm diameter green fluorescent microspheres in PBS (0.001% solids; Invitrogen) or CM-DiI were imaged through the cuticle. For dorsal viewing, mosquitoes were restrained on microscope slides using a non-invasive method previously used to measure circulatory physiology (Boppana and Hillyer, 2014). For lateral or ventral viewing, mosquitoes were positioned on their side or back. A real-time video was acquired under epi-fluorescence illumination on a Nikon SMZ1500 stereomicroscope connected to a Hamamatsu ORCA-Flash 2.8 digital CMOS camera (Hamamatsu Photonics, Hamamatsu, Japan) and NIS-Elements software, or the Nikon 90i compound microscope system described above. Microsphere-injected mosquitoes were immediately viewed or partially dissected to expose the aorta prior to viewing.

For brightfield video-imaging of the contracting aorta, mosquitoes were anesthetized on ice and the head, wings, a slice of the dorsal thorax, and the ventral side of the thorax and abdomen were removed by fine cutting using a blade. After dissection, mosquitoes were immediately placed dorsal side down on a glass slide containing PBS or further dissected to completely isolate the aorta. A real-time video was acquired on the Nikon SMZ1500 stereomicroscope system described above.

Tetracycline-resistant, green fluorescent protein (GFP)-expressing Escherichia coli (DH5 alpha) were grown as previously described (Coggins et al., 2012). Mosquitoes were anesthetized on ice, and E. coli were injected into the hemocoel as described above. Infections were performed with cultures at OD₆₀₀=5.0, and a plating assay revealed that the infection doses averaged 194,100 E. coli per mosquito. Two other groups of mosquitoes were maintained in parallel: one was injected with sterile LB broth (injury) and the other was left untreated (naïve).

Hemocytes were labeled with CM-DiI and the bulbous chamber, posterior aorta and conical chamber were isolated. Specimens were immediately visualized at 200–400× magnification, the hemocytes were counted, and still images were acquired using the Nikon 90i compound microscope. For a cell to be considered a hemocyte, it had to be 9–18 µm in diameter, and had to be labeled with both CM-DiI and Hoechst 33342. A minimum of seven mosquitoes were analysed for each treatment group (naïve, injury and infected). Data were analysed by ANOVA using Prism 6 Software (GraphPad, La Jolla, CA, USA).

In live A. gambiae, the contracting heart can be observed by intravital video-imaging of the abdomen, as this portion of the dorsal vessel extends along the dorsal midline and is tethered to the fairly translucent tergum by the alary muscles (Glenn et al., 2010; Estévez-Lao et al., 2013; League et al., 2015). At the thoraco-abdominal junction the heart connects to the aorta, which extends to the head. Because the thorax of A. gambiae is highly sclerotized, the aorta cannot be visualized by intravital imaging of intact mosquitoes. Furthermore, in dissected mosquitoes, the dense flight muscles of the thorax make identification of the aorta – which is also composed of muscle – difficult. To overcome these limitations, we examined the aorta using a combination of resection, fluorescence staining techniques, and still- and video-imaging of dissected and intact mosquitoes. Initially, the aorta was observed in dissected thoraces by tracing its trajectory as it extends from its junction with the heart towards the head (Fig. 1A,B). Muscle was stained with Alexa Fluor-conjugated phalloidin and the aorta was visualized by fluorescence microscopy in resected specimens (Fig. 1C).

The aorta originates in the head and extends posteriorly through the neck along the dorsal midline (Fig. 1B). In the head, the aorta associates with the dorsal plate of the pharyngeal pump (Fig. 2A–C), which is one of two pumps that facilitate feeding (Kim et al., 2011; Ha et al., 2015). Upon entering the prothorax, the aorta bends dorsally and then straightens such that it travels near the lateral midline with a slight, ventral slant as it crosses the mesothorax and metathorax (Fig. 1B). At a point near the thoraco-abdominal junction, the aorta bends sharply until it reaches the dorsal cuticle, and at the thoraco-abdominal junction it joins a region of the heart called the conical chamber.

In relation to other tissues, the anterior portion of the aorta is dorsal to the pharyngeal plate (Fig. 2A–C), which is also where the recurrent nerve passes between the lobes of the brain (Christophers, 1960). Along its path, the aorta lies just dorsal to the alimentary tract: in the head and neck it is dorsal to the foregut (pharynx and esophagus), and in the thorax it is dorsal to the midgut, the crop and the junction between the gastric caeca and the midgut (Fig. 2D–G). The gastric caeca are diverticula located at the junction of the foregut and midgut (Jones and Zeve, 1968; Okech et al., 2008), and although in the prothorax they diverge ventral of the aorta, they extend dorsally such that the pair straddle the aorta laterally (Fig. 2D–G). The corpora allata, which are a pair of neuroendocrine glands located in the prothorax, lie alongside the aorta (Fig. 2H), and the salivary glands are distal, as they lie ventral to the alimentary tract (Fig. 2F).

Like the heart, the aorta is a muscular tube formed by striated muscle arranged in a helical twist (Fig. 3A–F). The diameter of the aorta varies along its length but on average is narrower than the heart (Figs 1C and 3A). Based on both the diameter of the vessel and the arrangement of muscle fibers, the aorta is anatomically formed of three contiguous but distinct regions: the anterior aorta, the bulbous chamber and the posterior aorta (Fig. 3). At the anteriormost portion of the aorta – in the head – is an excurrent opening from which hemolymph exits the aorta and enters the hemocoel (Fig. 3B).

The anterior portion of the aorta originates in the head and extends into the neck. It makes up a little less than a quarter of the total length of the aorta and begins with an excurrent opening located above the dorsal pharyngeal plate of the head (Fig. 3B). The area of the excurrent opening resembles a prolate spheroid with an opening in the anterior end. Posterior to this structure, the aorta narrows significantly and consists of muscle fibers arranged in both right- and left-handed helical twists, with discontinuous longitudinal muscles (Fig. 4A). The cardiomyocyte nuclei are paired along the length of the anterior aorta (Fig. 4B,C). Compared with the heart, the density of muscle fibers is higher in the anterior aorta. The anterior aorta ends where the aorta bends dorsally in the prothorax, and the transition between the anterior aorta and the bulbous chamber typically contains a cluster of cardiomyocyte nuclei (Fig. 4C).

Immediately posterior of the anterior aorta is the bulbous chamber. It makes up just over a quarter of the aorta. The bulbous chamber begins at the dorsal bend in the prothorax and extends into the mesothorax (Figs 3A and 5A). The bulbous chamber is the region of the aorta with the densest muscle arrangement, and relative to the anterior aorta the cardiomyocytes are closer together (Figs 3A and 4C). As in the anterior aorta, the muscle arrangement in the bulbous chamber is in both right- and left-handed helical twists (Fig. 5B), but the diameter of the aorta gradually widens as it transitions from the anterior aorta into the bulbous chamber. Thin muscle fibers attach to the bulbous chamber and these fibers extend through the flight muscles and presumably tether the aorta to the cuticle (Figs 1A and 5C). Fat body is often adhered to the bulbous chamber, with most of it associated with the sites where the thin muscle fibers adhere to the aorta (Fig. 5C–E). The end of the bulbous chamber is formed of a cluster of cardiomyocyte nuclei, and the posterior aorta begins where the aorta narrows and the density of muscle decreases (Fig. 3E).

The posterior aorta constitutes approximately half of the aorta and extends from the middle of the mesothorax to the thoraco-abdominal junction, where it joins the heart. At the transition between the bulbous chamber and the posterior aorta the muscle changes from the densest arrangement to the sparsest arrangement, and this is evident by both the intensity of muscle staining and the increased spacing between the cardiomyocyte nuclei (Figs 4C and 6A). The muscle fibers in the posterior aorta are arranged in a left-handed helical twist, which is different from the overlapping left- and right-handed twists present in the anterior aorta and the bulbous chamber (Fig. 6B). As the posterior aorta approaches the heart, the muscle density modestly increases and the cardiomyocyte spacing decreases to levels similar to the heart (Fig. 4C). The posterior aorta widens at its end where the conical chamber of the heart begins (Fig. 6C).

The aorta terminates at the thoraco-abdominal junction where it joins a diamond-shaped region of the heart called the ‘conical chamber’ (Fig. 7A–C). The conical chamber begins where the musculature transitions into overlapping left- and right-handed twists, as in the anterior aorta and bulbous chamber (Fig. 7B). The shape of the conical chamber is present in both intact (Fig. 7C) and resected specimens (Figs 6C and 7A), indicating that it is not solely maintained by the tension of the alary muscles. The posterior aorta and the conical chamber are both anchored to the cuticle by an incomplete pair of alary muscles that join the dorsal vessel near where the anteriormost portion of the first abdominal tergite meets the postnotum. About a quarter of the alary muscle bundle associates with the posterior aorta, whereas the other three-quarters attaches to the conical chamber (Fig. 7C,D). The posterior myofibers of these alary muscles extend along the dorsal vessel such that they meet the alary muscles from the second abdominal segment (Fig. 7C). On abdominal segments 2–7, pericardial cells associate with the alary muscles of the heart (Glenn et al., 2010; Martins et al., 2011; King and Hillyer, 2012). We discovered that pericardial cells also associate with the alary muscles that attach to the posterior aorta and conical chamber (Fig. 7D,E). Compared with the pericardial cells present in abdominal segments 2–7, these cells are similar in terms of size, their binucleated nature, and their flanking of the dorsal vessel.

In the conical chamber is the heart's first ostial pair, called the thoraco-abdominal ostia (Fig. 7F). Like the ostia in abdominal segments 2–7, this ostial pair is located slightly posterior of the anterior suture of the first abdominal segment, but with this being the shortest segment, this ostial pair is only slightly anterior of the transverse midline instead of in the anterior portion of the segment. Thus, these ostia are positioned in the posterior half of the conical chamber – near where the posterior two bundles of the alary muscle attach to the heart – and the funnel-shaped lips of the ostia are angled towards the posterior of the body (Fig. 7E,F).

To determine the functional mechanics of hemolymph flow at the thoraco-abdominal junction, we conducted intravital video-imaging of live mosquitoes and observed the movement of CM-DiI-labeled hemocytes and density-neutral fluorescent microspheres in the dorsal portion of the posterior thorax and the first abdominal segment. These regions contain the venous channels, the conical chamber and the thoraco-abdominal ostia. Intravital imaging revealed that during retrograde heart contractions, hemocytes and fluorescent microspheres flowing through the venous channels as well as the dorsal portion of the first abdominal segment enter the heart through the thoraco-abdominal ostia and are propelled towards the posterior of the body (Movie 1). When the heart switches from contracting retrograde to contracting anterograde, hemolymph flow through the venous channels continues momentarily and then pauses or slows (Movie 1). During the period of pause or slowing, this hemolymph oscillates between moving anterograde and retrograde, but there is a small, net displacement towards the thoraco-abdominal ostia (Movie 1). Also, during anterograde heart contractions the hemolymph in the dorsal portion of the first abdominal segment flows away from the heart, whereas during retrograde heart contractions it flows towards the thoraco-abdominal ostia.

Based on these observations we infer that during retrograde heart contractions the thoraco-abdominal ostia receive hemolymph from the venous channels and the abdominal hemocoel of the first segment, and once in the lumen of the dorsal vessel this hemolymph is propelled towards the posterior of the body. During anterograde heart contractions, the thoraco-abdominal ostia close, and thus hemolymph does not enter the lumen of the dorsal vessel. Instead, the retrograde movement of hemolymph travelling through the venous channels significantly slows, and this hemolymph flows past the thoraco-abdominal ostia and into the first abdominal segment, causing hemolymph in this segment to flow slowly away from the heart.

To observe the aorta in live specimens, mosquitoes were anesthetized, the ventral side of the thorax was removed in a manner that exposed the aorta, and the tissues were visualized by intravital video-imaging. Visualization of the aorta while in situ revealed that it pulsates: the diameter expands and retracts (Movie 2). Because of challenges associated with this procedure we were only able to measure the aortic contraction rate in a limited number of mosquitoes. In those mosquitoes, where the aorta remained within the indirect flight muscles, the aortic contraction rate was 1.5–2.0 Hz, similar to the 1.3–2.2 Hz contraction rate we typically observe for the heart (Glenn et al., 2010; Estévez-Lao et al., 2013; League et al., 2015). These contractions appear different from the contractions of the heart. Whereas the heart contracts in a smooth, wave-like manner and alternates between contracting in the anterograde and retrograde directions (Glenn et al., 2010; League et al., 2015), the aorta contracts in two phases that result in the shortening of distinct regions of the vessel (Movie 2). Specifically, in the first phase the posterior aorta shortens, and this is followed by a second phase where the bulbous chamber shortens. The contraction of the bulbous chamber is significantly more vigorous than the contraction of the posterior aorta, and the sequence of these contractions is suggestive of net displacement in the anterograde direction (Movie 2). On occasions, one of the pulsations occurred without the other, but we hypothesize that this was because of the disruption associated with the dissection of the mosquito.

To determine whether aortic contractions are dependent on the heart, we isolated the aorta and examined its contraction dynamics. When the aorta was attached to the anterior half of the heart, the heart did not contract, yet the aorta contracted (Movie 2). The majority of contractions originated from the bulbous chamber. When the aorta was isolated from the heart, contractions still occurred (Movie 2). Together, these data strongly suggest that contractions of the bulbous chamber are not dependent on the wave-like contractions of the heart.

Although the aorta cannot be directly visualized in the thorax or head of intact mosquitoes, hemolymph released into the head via the anterior excurrent opening of the aorta has to flow in the retrograde direction through the hemocoel of the neck as it returns to the thorax and abdomen. To determine whether hemolymph flow through the aorta is continuous, we used intravital video-imaging to examine flow within the neck of intact mosquitoes. In the neck, hemolymph only flows in the retrograde direction, but this flow periodically pauses for about 5 s and does so approximately six times per minute (Movie 3). These pauses roughly correspond to the number of heartbeat directional reversals, and the time the heart spends contracting in the retrograde direction (Glenn et al., 2010; Estévez-Lao et al., 2013; League et al., 2015). During periods of pause, hemolymph in the neck oscillates back and forth but there is no net displacement. This indicates that hemolymph flows through the aorta only in the anterograde direction, and does so only during periods of anterograde heart contractions. This also indicates that the pulsations of the aorta – which are continuous – do not propel hemolymph during periods of retrograde heart contractions.

Hemocytes are always present at the periostial regions of the heart, which are the locations that flank the incurrent ostia of abdominal segments 2–7 (King and Hillyer, 2012; Sigle and Hillyer, 2016). Upon infection, additional hemocytes aggregate in these regions of high hemolymph flow, where they phagocytose and kill pathogens (King and Hillyer, 2012; Sigle and Hillyer, 2016). To determine whether hemocytes are also attached to the aorta, the hemocytes of live mosquitoes were labeled with CM-DiI, the aorta was isolated, and fluorescence microscopy was used to visualize these cells. In some but not all mosquitoes, hemocytes were indeed present on the surface of the aorta. When present, hemocytes were very rarely seen in the anterior aorta. Instead, they were predominantly in the bulbous chamber and occasionally on the posterior aorta, but their number was always very low. For example, uninfected mosquitoes usually had fewer than two hemocytes (Fig. 8A–C).

To determine whether infection induces the aggregation of hemocytes on the surface of the aorta, mosquitoes were injured, infected with E. coli or left unmanipulated, and the number of hemocytes present on the bulbous chamber and posterior aorta counted 24 h later. Overall, injured or E. coli-infected mosquitoes had two to three hemocytes on the aorta, which is similar to that observed in naïve mosquitoes (Fig. 8A–E; ANOVA, P=0.3756). Some mosquitoes, regardless of treatment, lacked hemocytes on the aorta. Upon infection, hemocytes on the aorta phagocytosed GFP-expressing E. coli, indicating that they are immunologically active (Fig. 8D,E). Occasionally, melanization was observed on the surface of the bulbous chamber, but melanin deposits were associated with fat body and not the hemocytes (Fig. 8F).

In this study we used both ex vivo and in vivo imaging to characterize the structure and function of the aorta and conical chamber of adult A. gambiae. Furthermore, we describe how alary muscles, pericardial cells and hemocytes attach to these tissues, and we illustrate how hemolymph flows through the aorta and the conical chamber, including how it flows through the thoraco-abdominal ostia (Fig. 9).

Holometabolous insects undergo drastic morphological and physiological changes during their development, and in some the structure of the dorsal vessel varies greatly between life stages whereas in others it stays largely the same (Jones, 1977). In mosquitoes, the heart of larvae and adults is structurally similar (Leódido et al., 2013; League et al., 2015). However, in A. gambiae the adult heart contracts in both anterograde and retrograde directions whereas the larval heart contracts only in an anterograde direction. Furthermore, hemolymph enters the adult heart through ostia located in abdominal segments 2–7, but in larvae the abdominal ostia are always closed (Glenn et al., 2010; League et al., 2015). Instead, hemolymph enters the larval heart through a pair of incurrent openings located at the posterior of the body – structures that in adults only have excurrent function (League et al., 2015). Comparison of the aorta of larvae and adults reveals that they are structurally similar, but that there are minor differences in the path the aorta takes. In A. gambiae larvae and adults, the aorta extends from the head to the thorax in a similar fashion, with the anterior portion of the bulbous chamber slanting dorsally as it expands in diameter (League et al., 2015). In larvae, hemolymph flowing in an anterograde direction through the aorta exits into the hemocoel in the posterior region of the head (League et al., 2015), which is the same location where hemolymph exits the aorta of adults. However, the path of the posterior aorta differs between larvae and adults; in larvae it extends along a linear plane, whereas in adults it bends dorsally to reach the heart. These differences reflect differences in the adult and larval body plans: the larval thorax and abdomen are more streamlined when compared with the adult.

The aorta has been cursorily described in other mosquito species (Jones, 1952, 1954; Christophers, 1960). These studies did not publish micrographs – making it difficult or impossible to assess and compare structures – and did not visualize the flow of hemolymph. Comparing the structure of the aorta of A. gambiae to what has been described of the aorta of larvae, pupa and adults of Aedes aegypti and Anopheles quadrimaculatus reveals similarities in structure and general location (Jones, 1952, 1954; Christophers, 1960). These include a smaller lumen compared with the lumen of the heart, a bulbous-like sinus in the prothorax, an excurrent opening in the head, and a lack of ostia. The excurrent opening of the aorta of A. quadrimaculatus and A. aegypti larvae and adults has been described as being located between the brain and the pharynx (Jones, 1954; Christophers, 1960). Furthermore, in A. aegypti adults the excurrent opening is described as ‘semicircular in section with a flat floor where it lies upon the narrow anterior portion of the dorsal plate of the oesophageal pump’ (Christophers, 1960), similar to our observations in resected specimens. Altogether these data suggest that the aorta terminates above the dorsal pharyngeal plate. However, earlier studies provided conflicting statements on whether the aorta extends beyond the anterior of the pharynx (Clements, 1956; Christophers, 1960). In the present study, we observed the aorta terminating as a prolate spheroid immediately above the dorsal pharyngeal plate. Although we did not observe additional musculature extending anterior of this region, if such muscle were to be present it would serve as attachment muscle to the anterior of the head and as a means of supporting the antennal hearts (Clements, 1956; Sun and Schmidt, 1997; Boppana and Hillyer, 2014). In summary, comparison of our data with earlier cursory descriptions of the aorta of other mosquito species suggests that the function of this region of the dorsal vessel is conserved across the mosquito lineage (Jones, 1952, 1954). Comparison to the aorta of adult Drosophila is difficult because the only description does not include micrographs (Miller, 1950); however, the general location of the aorta appears to be conserved.

Although the aorta has often been described as a vessel that does not contract and merely transports hemolymph propelled by the heart (Klowden, 2013), contraction of the aorta has been reported in some insects, including the adult blowfly Phormia regina, the mealworm beetle Tenebrio molitor, the Vietnamese stick insect Baculum extradentatum, the larval fruit fly Drosophila melanogaster, the locust Locusta migratoria and multiple Lepidopteran species (Hessel, 1969; Angioy and Pietra, 1995; Markou and Theophilidis, 2000; Hertel and Pass, 2002; Dasari and Cooper, 2006; Ejaz and Lange, 2008; da Silva et al., 2011). In some of these insects the contractions are intrinsic, as contractions continue after the aorta has been separated from the heart (Dasari and Cooper, 2006; Ejaz and Lange, 2008; da Silva et al., 2011). This is similar to what we observed in A. gambiae. Furthermore, the insect heart contracts myogenically, but the contraction rate and the proportional directionality is influenced by neurotransmitters and neurohormones (Hillyer, 2015). Although nerve processes have been described along the wall of the aorta in Baculum extradentatum and Drosophila larvae (Johnstone and Cooper, 2006; Ejaz and Lange, 2008), innervation of the mosquito heart has only been cursorily described (Estévez-Lao et al., 2013). We hypothesize that, like the heart, aortic contractions are myogenic. Although the aorta contracts continuously, these contractions appear to play a minor role in propelling hemolymph to the head during anterograde heart contractions. Instead, these contractions, together with the narrow nature of the aorta, may maintain pressure during periods of anterograde heart flow and prevent backflow during periods of retrograde heart flow.

A lack of alary muscles is often used as a defining feature differentiating the aorta in the thorax from the heart in the abdomen (Clements, 1992; Nation, 2008); however, we show that there is an incomplete alary muscle pair that originates in the thorax and supports the posterior of the aorta as it bends sharply as it approaches the thoraco-abdominal junction and joins the conical chamber. Also, thin muscle fibers anchor the bulbous chamber in the prothorax, which could be an additional, delicate alary muscle pair. This, and previous reports of alary muscle associating with the aorta of an insect (Ejaz and Lange, 2008), suggests that varying body plans require different types of aortic muscle support, and these muscles may be necessary for the maintenance of shape during contractions. Broadly, the aorta and heart differ significantly in muscle structure and contractility, and while variations in the presence or absence of muscles supporting the aorta exist, when present these muscles are less robust in comparison with those of the heart. As with the alary muscles of the heart, the alary muscles of the posterior aorta contain pericardial cells. This is similar to Culex fatigans, where pericardial cells are present at the posterior border of the metathorax (Pal, 1944).

The heart of adult mosquitoes periodically alternates between contracting in the anterograde and retrograde directions (Glenn et al., 2010; League et al., 2015). Early in adulthood the majority of heart contractions propagate in the anterograde direction, whereas later in life the proportion of contractions is evenly split between anterograde and retrograde (Doran et al., 2017). In this study, we report that during retrograde heart contractions, hemolymph in the venous channels and first abdominal segment enters the conical chamber through the thoraco-abdominal ostia, and flows in a retrograde direction via the lumen of the heart. During anterograde heart contractions, however, the thoraco-abdominal ostia close and do not allow the passage of hemolymph. This is opposite from the incurrent ostia of abdominal segments 2–7, which in mosquitoes are only open during anterograde heart contractions. This is also different from Drosophila species, where the thoraco-abdominal ostia have been described as either accepting hemolymph into the dorsal vessel during both anterograde and retrograde heart contractions (Wasserthal, 2007), or allowing inflow of hemolymph during retrograde heart contractions and outflow during the latter periods of anterograde heart contractions (Sláma, 2010). Furthermore, in adult Drosophila species and Calliphora vicina, two ostial pairs are present within the conical chamber: one in the first abdominal segment and one in the second (Wasserthal, 1999, 2007). In comparison, the conical chamber in A. gambiae does not extend into the second abdominal segment and only contains one pair of ostia: the thoraco-abdominal ostia. Finally, we did not detect any ostia on the aorta. In Drosophila larvae, cells expressing seven up (svp) give rise to the ostia of the heart; however, svp-expressing cells in the aorta do not differentiate into functional ostia (Ponzielli et al., 2002). The muscle formations that associate with these svp-expressing cells is reminiscent of what we observe in the bulbous chamber and anterior aorta of adult A. gambiae.

Hemocytes are immune cells that circulate with the hemocoel or are attached to tissues (sessile). In adult mosquitoes, an infection induces the aggregation of hemocytes at the periostial regions of the heart (King and Hillyer, 2012, 2013). This is advantageous because it places immune cells in the areas of the hemocoel with the highest hemolymph flow, thus allowing for the sequestration and killing of pathogens (Glenn et al., 2010; King and Hillyer, 2012; Sigle and Hillyer, 2016). In mosquito larvae, however, hemocytes do not aggregate at the periostial regions of the heart and instead aggregate in the tracheal tufts (League and Hillyer, 2016). This occurs because of differences in hemolymph flow dynamics between larvae and adults; in larvae the abdominal ostia are always closed, and thus the periostial regions are not locations of high hemolymph flow (League et al., 2015). Instead, hemolymph enters the heart via a posterior excurrent opening that is enveloped by the tracheal tufts, and thus in larvae there is also a high density of hemocytes in the areas of the heart that experience the most flow. Here we assessed whether hemocytes attach to the exterior of the aorta. We observed that although hemocytes are present on the aorta of adult mosquitoes, there are usually fewer than five hemocytes even in an infected mosquito, which is significantly fewer than the >100 hemocytes that adhere to the periostial regions of the heart (King and Hillyer, 2012, 2013; Sigle and Hillyer, 2016). The lack of hemocyte aggregation on the aorta following an infection is not surprising. Ostia are lacking in this portion of the dorsal vessel. Thus, the external regions of the aorta do not experience high hemolymph flow.

Across Insecta there are significant differences in the structure of the aorta. This includes the presence or absence of chambers, diverticula, loops, folds and ostia (Jones, 1977). In the present study we characterize the structural mechanics and functional role of the aorta of adult A. gambiae, a mosquito of epidemiological relevance because of its ability to transmit malaria. This research expands our understanding of how hemolymph is circulated throughout the body, and how the circulatory and immune systems interact during the course of an infection.

We thank Tania Estévez-Lao for mosquito husbandry, discussions and critically commenting on this manuscript. We also thank Dr Garrett League for enlightening conversations.